Deep slate diagnosis — 20260912

• Games: 15 · hits 8 · misses 7 · winner accuracy 0.5333
• Mean total-run bias on losses: -1.286 (negative = under-projected)
• Mean team-run error on losses: 6.143 · mean Brier on losses: 0.302

Loss primary drivers

• Small edge on the losing side: 1
• Several errors of similar size: 1
• Run environment far higher than projected (under-dispersed): 1
• Starting-pitcher projection was wrong: 2
• Genuine toss-up that flipped: 1
• Backed the wrong side with real conviction: 1

High-level causal issue register

[CRITICAL] Published probability differs materially without a recorded transformation

• Domain: probability_lineage · affected: CHW@STL, COL@DET
• Evidence: 2 game(s) have an unexplained ≥2pp calibrated-to-published delta
• Action: Block any promotion until the missing probability transformation is recorded and reconciled.
• Do not infer: Do not assume the probability governor was bypassed until you have traced the pick-side math.

[HIGH] Published margins are compressed relative to realized margins

• Domain: score_distribution · affected: BAL@TOR, NYM@NYY, PHI@ATL, SEA@OAK, TEX@ARI
• Evidence: 12/15 predictions were ≤2 runs; 8/15 actual margins were ≥4; mean predicted margin 1.67 vs actual 4.33
• Action: Check whether predicted margins are too narrow across several recent out-of-sample slates (measuring tail coverage and margin MAE) before widening the score spread.
• Do not infer: A single blowout game is not a reason to widen every score prediction.

[HIGH] Published confidence buckets were poorly aligned on this slate

• Domain: calibration · affected: BAL@TOR, CHW@STL, CIN@MIL, CLE@MIN, COL@DET, HOU@TB, KC@BOS, LAA@WAS, LAD@MIA, NYM@NYY, PHI@ATL, PIT@CHC, SD@SF, SEA@OAK, TEX@ARI
• Evidence: weighted absolute calibration error 0.1711 across 15 games
• Action: Watch the same confidence buckets over the full rolling window before changing any calibration weight.
• Do not infer: A single slate has far too few games to establish a stable calibration curve.

[HIGH] Validated run models explain very little holdout variance

• Domain: model_validation · affected: BAL@TOR, CHW@STL, CIN@MIL, CLE@MIN, COL@DET, HOU@TB, KC@BOS, LAA@WAS, LAD@MIA, NYM@NYY, PHI@ATL, PIT@CHC, SD@SF, SEA@OAK, TEX@ARI
• Evidence: home/away/total R² = 0.0118/0.0210/0.0207
• Action: Treat exact run scores as low-confidence estimates, and look for stronger features or a better target before tuning the score tails.
• Do not infer: A small positive R² only means the model beats guessing the average; it does not make the exact scores reliable.

[INFO] Every settled game received structural QA

• Domain: coverage · affected: BAL@TOR, CHW@STL, CIN@MIL, CLE@MIN, COL@DET, HOU@TB, KC@BOS, LAA@WAS, LAD@MIA, NYM@NYY, PHI@ATL, PIT@CHC, SD@SF, SEA@OAK, TEX@ARI
• Evidence: 15/15 games structurally audited
• Action: Keep the per-game audit receipt in every report.
• Do not infer: Full audit coverage does not cancel the findings raised inside the audited games.

[INFO] Simulation produced no winner-direction intervention on this slate

• Domain: simulation · affected: BAL@TOR, CHW@STL, CIN@MIL, CLE@MIN, COL@DET, HOU@TB, KC@BOS, LAA@WAS, LAD@MIA, NYM@NYY, PHI@ATL, PIT@CHC, SD@SF, SEA@OAK, TEX@ARI
• Evidence: flip candidates 0; published flips 0
• Action: Measure the simulation's contribution across rolling slates, separating score realization from winner flipping.
• Do not infer: Zero winner flips is not a defect and does not mean the simulation did no score-reconciliation work.

Cross-slate trends (persistent-pattern memory)

• Window: 20260902, 20260903, 20260904, 20260905, 20260906, 20260907, 20260908, 20260909, 20260910, 20260911, 20260912 (10 prior slates loaded)
• Highest-confidence pick loss streak: 2 slate(s)
• Actionable pattern detected: True

Escalated persistent patterns

• simulation_inert_on_slate — 11 consecutive slates (11 in window)
• Target lever(s): simulation.allow_score
• Direction: confirm the simulation module is running, not silently no-oping
• Guardrail: simulation_inert_on_slate has persisted 11 consecutive slates; this is no longer single-slate noise. Treat it as an actionable pattern and raise the code/diagnostic action named; do not force a parameter move.
• weak_run_model_signal — 11 consecutive slates (11 in window)
• Target lever(s): training.xgb.n_estimators, training.xgb.max_depth, training.xgb.learning_rate
• Direction: investigate features/target via isolated training backtest (manual)
• Guardrail: weak_run_model_signal has persisted 11 consecutive slates; this is no longer single-slate noise. Treat it as an actionable pattern and raise the code/diagnostic action named; do not force a parameter move.

Metric direction across the window

• winner_accuracy: regressing · series [0.3333, 0.5556, 0.625, 0.5333, 0.6, 0.5455, 0.7333, 0.6667, 0.6, 0.6667, 0.5333]
• weighted_calibration_error: regressing · series [0.21, 0.1474, 0.1316, 0.1, 0.0937, 0.2611, 0.1825, 0.1069, 0.1171, 0.2091, 0.1711]
• margin_bias: regressing · series [-3.067, -1.778, -0.688, -1.467, -1.467, -0.182, -0.867, -2.133, -1.0, -2.933, -2.667]
• mean_team_run_error_on_losses: regressing · series [6.4, 6.75, 6.0, 5.286, 5.333, 3.8, 6.75, 5.6, 6.5, 5.0, 6.143]

Pipeline stage quality

• historical_ml: winner 8/15 (0.5333), team-run MAE 2.8836, margin MAE 4.202
• variance_applied: winner 7/15 (0.4667), team-run MAE 2.8333, margin MAE 4.2
• enhanced_prediction: winner 8/15 (0.5333), team-run MAE 3.0333, margin MAE 4.2
• embedded_savant_run_environment: winner 8/15 (0.5333), team-run MAE 3.0, margin MAE 4.1333
• fundamental_edge_target_review: winner 8/15 (0.5333), team-run MAE 3.0, margin MAE 4.1333
• capped_target: winner 8/15 (0.5333), team-run MAE 3.0, margin MAE 4.1333
• simulation_result: winner 8/15 (0.5333), team-run MAE 3.0, margin MAE 4.1333
• published_final: winner 8/15 (0.5333), team-run MAE 3.0, margin MAE 4.1333

Probability lineage

• BAL@TOR: raw 0.646 → calibrated 0.5871 → published 0.53; score coherence followed by matchup-recency guardrail
• CHW@STL: raw 0.6585 → calibrated 0.6339 → published 0.5908; no recorded transformation explains the publication delta
• CIN@MIL: raw 0.7249 → calibrated 0.7235 → published 0.54; score coherence followed by matchup-recency guardrail
• CLE@MIN: raw 0.5951 → calibrated 0.5578 → published 0.5412; score coherence followed by matchup-recency guardrail
• COL@DET: raw 0.7568 → calibrated 0.7296 → published 0.5989; no recorded transformation explains the publication delta
• HOU@TB: raw 0.5702 → calibrated 0.5437 → published 0.5315; score coherence followed by matchup-recency guardrail
• KC@BOS: raw 0.5915 → calibrated 0.5523 → published 0.53; score coherence followed by matchup-recency guardrail
• LAA@WAS: raw 0.5 → calibrated 0.5078 → published 0.5056; published probability matches the calibrated/governed value
• LAD@MIA: raw 0.6979 → calibrated 0.6956 → published 0.6; score coherence followed by matchup-recency guardrail
• NYM@NYY: raw 0.5478 → calibrated 0.5392 → published 0.5392; published probability matches the calibrated/governed value
• PHI@ATL: raw 0.5 → calibrated 0.5065 → published 0.505; published probability matches the calibrated/governed value
• PIT@CHC: raw 0.5 → calibrated 0.5547 → published 0.53; score coherence followed by matchup-recency guardrail
• SD@SF: raw 0.5 → calibrated 0.5078 → published 0.505; published probability matches the calibrated/governed value
• SEA@OAK: raw 0.5574 → calibrated 0.5346 → published 0.525; score coherence followed by matchup-recency guardrail
• TEX@ARI: raw 0.5911 → calibrated 0.5834 → published 0.5589; score coherence followed by matchup-recency guardrail

Every game, in detail

BAL@TOR — ✅ hit

• Predicted TOR 3-4 (conf +0.06, lean) · actual TOR 3-7
• Root cause: solid_correct_call
• Errors: winner correct · Brier 0.2209 · team-run err 3.0 · total bias -3
• Contributing factors:
• total runs under-projected by 3 (7 vs 10)
• TOR offense under-projected by 3

CHW@STL — ❌ miss

• Predicted STL 3-4 (conf +0.18, moderate) · actual CHW 6-5
• Root cause: Backed the wrong side with real conviction
• Secondary cause: Small edge on the losing side
• Errors: winner WRONG · Brier 0.349039 · team-run err 4.0 · total bias -4
• Contributing factors:
• total runs under-projected by 4 (7 vs 11)
• CHW offense under-projected by 3
• committed to STL at 0.18 confidence and it lost

CIN@MIL — ✅ hit

• Predicted MIL 6-10 (conf +0.08, lean) · actual MIL 9-13
• Root cause: right_winner_wrong_score
• Errors: winner correct · Brier 0.2116 · team-run err 6.0 · total bias -6
• Contributing factors:
• total runs under-projected by 6 (16 vs 22)
• CIN offense under-projected by 3
• MIL offense under-projected by 3
• Shane Drohan (MIL) allowed 3 more ER than projected (5 vs 2)

CLE@MIN — ❌ miss

• Predicted CLE 6-4 (conf -0.08, lean) · actual MIN 3-4
• Root cause: Several errors of similar size
• Errors: winner WRONG · Brier 0.292863 · team-run err 3.0 · total bias +3
• Contributing factors:
• total runs over-projected by 3 (10 vs 7)
• CLE offense over-projected by 3
• Daniel Espino (CLE) allowed 3 fewer ER than projected (0 vs 3)

COL@DET — ✅ hit

• Predicted DET 4-7 (conf +0.20, moderate) · actual DET 7-11
• Root cause: right_winner_wrong_score
• Errors: winner correct · Brier 0.160881 · team-run err 7.0 · total bias -7
• Contributing factors:
• total runs under-projected by 7 (11 vs 18)
• COL offense under-projected by 3
• DET offense under-projected by 4

HOU@TB — ✅ hit

• Predicted TB 4-5 (conf +0.06, lean) · actual TB 2-3
• Root cause: solid_correct_call
• Errors: winner correct · Brier 0.219447 · team-run err 4.0 · total bias +4
• Contributing factors:
• total runs over-projected by 4 (9 vs 5)

KC@BOS — ✅ hit

• Predicted BOS 4-6 (conf +0.06, lean) · actual BOS 1-5
• Root cause: solid_correct_call
• Errors: winner correct · Brier 0.2209 · team-run err 4.0 · total bias +4
• Contributing factors:
• total runs over-projected by 4 (10 vs 6)
• KC offense over-projected by 3

LAA@WAS — ❌ miss

• Predicted LAA 4-3 (conf -0.01, toss_up) · actual WAS 5-6
• Root cause: Genuine toss-up that flipped
• Secondary cause: Small edge on the losing side
• Errors: winner WRONG · Brier 0.25568 · team-run err 4.0 · total bias -4
• Contributing factors:
• total runs under-projected by 4 (7 vs 11)
• WAS offense under-projected by 3
• Walbert Ureña (LAA) allowed 3 more ER than projected (4 vs 1)

LAD@MIA — ❌ miss

• Predicted LAD 7-4 (conf -0.20, strong) · actual MIA 3-4
• Root cause: Starting-pitcher projection was wrong
• Secondary cause: Backed the wrong side with real conviction
• Errors: winner WRONG · Brier 0.36 · team-run err 4.0 · total bias +4
• Contributing factors:
• total runs over-projected by 4 (11 vs 7)
• LAD offense over-projected by 4
• Tyler Glasnow (LAD) allowed 4 more ER than projected (4 vs 0)
• committed to LAD at 0.20 confidence and it lost

NYM@NYY — ❌ miss

• Predicted NYY 5-7 (conf +0.08, lean) · actual NYM 12-2
• Root cause: Small edge on the losing side
• Errors: winner WRONG · Brier 0.290715 · team-run err 12.0 · total bias -2
• Contributing factors:
• NYM offense under-projected by 7
• NYY offense over-projected by 5

PHI@ATL — ❌ miss

• Predicted PHI 4-3 (conf -0.01, toss_up) · actual ATL 2-12
• Root cause: Run environment far higher than projected (under-dispersed)
• Secondary cause: Genuine toss-up that flipped
• Errors: winner WRONG · Brier 0.255025 · team-run err 11.0 · total bias -7
• Contributing factors:
• total runs under-projected by 7 (7 vs 14)
• ATL offense under-projected by 9

PIT@CHC — ✅ hit

• Predicted CHC 4-5 (conf +0.06, lean) · actual CHC 3-4
• Root cause: solid_correct_call
• Errors: winner correct · Brier 0.2209 · team-run err 2.0 · total bias +2

SD@SF — ✅ hit

• Predicted SD 5-4 (conf -0.01, toss_up) · actual SD 7-6
• Root cause: toss_up_went_our_way
• Errors: winner correct · Brier 0.245025 · team-run err 4.0 · total bias -4
• Contributing factors:
• total runs under-projected by 4 (9 vs 13)
• Cesar Perdomo (SF) allowed 3 more ER than projected (5 vs 2)

SEA@OAK — ✅ hit

• Predicted SEA 6-5 (conf -0.05, lean) · actual SEA 19-1
• Root cause: right_winner_wrong_score
• Errors: winner correct · Brier 0.22565 · team-run err 17.0 · total bias -9
• Contributing factors:
• total runs under-projected by 9 (11 vs 20)
• SEA offense under-projected by 13
• OAK offense over-projected by 4

TEX@ARI — ❌ miss

• Predicted ARI 4-5 (conf +0.12, moderate) · actual TEX 6-2
• Root cause: Starting-pitcher projection was wrong
• Secondary cause: Backed the wrong side with real conviction
• Errors: winner WRONG · Brier 0.312417 · team-run err 5.0 · total bias +1
• Contributing factors:
• ARI offense over-projected by 3
• Kumar Rocker (TEX) allowed 4 fewer ER than projected (1 vs 5)
• committed to ARI at 0.12 confidence and it lost

Cross-game lever tension

Each candidate fix, with the games it would help and the games it would hurt. A lever that repairs one game and breaks two is visible here before it is proposed.

score.variance_multiplier / score.variance_floor (widen run distributions)

• Targets: brier_score, calibration_error
• Net: 3 helped vs 1 hurt
• Would help:
• CHW@STL (miss): lost a 0.18-confidence pick; wider tails lower overconfidence
• LAD@MIA (miss): lost a 0.20-confidence pick; wider tails lower overconfidence
• TEX@ARI (miss): lost a 0.12-confidence pick; wider tails lower overconfidence
• Would hurt / risk:
• COL@DET (hit): strong correct call at 0.20; widening softens a good edge

player.pitcher.v4 form weights / player.pitcher_prior_games

• Targets: starter_er_mae, starter_outs_mae
• Net: 6 helped vs 3 hurt
• Would help:
• CIN@MIL (hit): starter ER off by 3
• CLE@MIN (miss): starter ER off by 3
• LAA@WAS (miss): starter ER off by 3
• LAD@MIA (miss): starter ER off by 4
• SD@SF (hit): starter ER off by 3
• TEX@ARI (miss): starter ER off by 4
• Would hurt / risk:
• CHW@STL (miss): starter projection was accurate (ER off 1); reweight risks disturbing it
• COL@DET (hit): starter projection was accurate (ER off 1); reweight risks disturbing it
• PIT@CHC (hit): starter projection was accurate (ER off 1); reweight risks disturbing it

recent. / season. run-mean weighting

• Targets: team_run_mae, total_run_mae, accuracy
• Net: 12 helped vs 1 hurt
• Would help:
• CHW@STL (miss): team runs off by 4
• CIN@MIL (hit): team runs off by 6
• COL@DET (hit): team runs off by 7
• HOU@TB (hit): team runs off by 4
• KC@BOS (hit): team runs off by 4
• LAA@WAS (miss): team runs off by 4
• LAD@MIA (miss): team runs off by 4
• NYM@NYY (miss): team runs off by 12
• PHI@ATL (miss): team runs off by 11
• SD@SF (hit): team runs off by 4
• SEA@OAK (hit): team runs off by 17
• TEX@ARI (miss): team runs off by 5
• Would hurt / risk:
• PIT@CHC (hit): team runs already close (off 2); reweight risks disturbing it